{
  "id": 76979,
  "title": "Problems with kernels ",
  "url": "/competitions/quora-insincere-questions-classification/discussion/76979",
  "author_name": "",
  "post_date": "2019-01-08T12:41:31.366000",
  "votes": 3,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Hey. </p>\n\n<p>I could not make a submission for the last three days. I run the exact same kernel that worked a week ago, but now I receive an error: </p>\n\n<p>... I could not publish the post with a problem, I dont know why. See attached picture ... </p>\n\n<p>Also, the time run of kernel increased from 6500s up to 15000s. I've heard that those problems might be a result of a massive kernels update. Is it so? Does anyone have a similar problem as I do? </p>\n\n<p>PS: I tried to post this message four times in a row, but only one was sucsessful. What is happening with kaggle? </p>",
  "messages": [
    {
      "id": 452249,
      "postDate": "2019-01-08T12:41:31.367Z",
      "content": "<p>Hey. </p>\n\n<p>I could not make a submission for the last three days. I run the exact same kernel that worked a week ago, but now I receive an error: </p>\n\n<p>... I could not publish the post with a problem, I dont know why. See attached picture ... </p>\n\n<p>Also, the time run of kernel increased from 6500s up to 15000s. I've heard that those problems might be a result of a massive kernels update. Is it so? Does anyone have a similar problem as I do? </p>\n\n<p>PS: I tried to post this message four times in a row, but only one was sucsessful. What is happening with kaggle? </p>",
      "rawMarkdown": "Hey. \n\nI could not make a submission for the last three days. I run the exact same kernel that worked a week ago, but now I receive an error: \n\n... I could not publish the post with a problem, I dont know why. See attached picture ... \n      \nAlso, the time run of kernel increased from 6500s up to 15000s. I've heard that those problems might be a result of a massive kernels update. Is it so? Does anyone have a similar problem as I do? \n\nPS: I tried to post this message four times in a row, but only one was sucsessful. What is happening with kaggle? ",
      "votes": 3
    },
    {
      "id": 453291,
      "postDate": "2019-01-10T01:10:05.993Z",
      "content": "<p>Hi @DenisVorotyntsev</p>\n\n<p>From the logs you attached, it seems like you are using a Kernel with GPU disabled to train a CudnnRNN model which requires a GPU. </p>\n\n<p>Can you confirm that the GPU setting in the Kaggle Kernel Editor is turned on for you kernel?</p>\n\n<p>Thank you</p>",
      "rawMarkdown": "Hi @DenisVorotyntsev\n\nFrom the logs you attached, it seems like you are using a Kernel with GPU disabled to train a CudnnRNN model which requires a GPU. \n\nCan you confirm that the GPU setting in the Kaggle Kernel Editor is turned on for you kernel?\n\nThank you",
      "votes": 1,
      "replies": [
        {
          "id": 453444,
          "postDate": "2019-01-10T07:56:10.277Z",
          "content": "<p>Hi Vincent, </p>\n\n<p>Thank you for reply. GPU was off. I turned it on. </p>",
          "rawMarkdown": "Hi Vincent, \n\nThank you for reply. GPU was off. I turned it on. "
        },
        {
          "id": 454440,
          "postDate": "2019-01-11T16:38:10.813Z",
          "content": "<p>Hi <a href=\"/huytran\">@huytran</a> &amp; <a href=\"/sezaugg\">@sezaugg</a>,</p>\n\n<p>Does making sure the GPU is turned ON for your kernel also fixes the problem for you?</p>",
          "rawMarkdown": "Hi @huytran &amp; @sezaugg,\n\n Does making sure the GPU is turned ON for your kernel also fixes the problem for you?"
        }
      ]
    },
    {
      "id": 452296,
      "postDate": "2019-01-08T14:12:13.947Z",
      "content": "<p>Cudnn and tensorflow versions mismatch.</p>",
      "rawMarkdown": "Cudnn and tensorflow versions mismatch.",
      "votes": 1
    },
    {
      "id": 452260,
      "postDate": "2019-01-08T13:09:14.503Z",
      "content": "<p>This looks like a Tensorflow update.</p>",
      "rawMarkdown": "This looks like a Tensorflow update.",
      "votes": 1
    },
    {
      "id": 452380,
      "postDate": "2019-01-08T16:42:00.090Z",
      "content": "<p>I got the same issue with my new kernels. The old ones are still working fine.</p>",
      "rawMarkdown": "I got the same issue with my new kernels. The old ones are still working fine."
    },
    {
      "id": 452368,
      "postDate": "2019-01-08T16:20:11.617Z",
      "content": "<p>Sometimes fork an run the new kernel magically solves things. </p>",
      "rawMarkdown": "Sometimes fork an run the new kernel magically solves things. ",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 453291,
      "author_name": "Vincent Roseberry",
      "author_url": "",
      "post_date": "2019-01-10T01:10:05.993000",
      "content": "<p>Hi @DenisVorotyntsev</p>\n\n<p>From the logs you attached, it seems like you are using a Kernel with GPU disabled to train a CudnnRNN model which requires a GPU. </p>\n\n<p>Can you confirm that the GPU setting in the Kaggle Kernel Editor is turned on for you kernel?</p>\n\n<p>Thank you</p>",
      "votes": 1,
      "replies": [
        {
          "id": 453444,
          "author_name": "",
          "author_url": "",
          "post_date": "2019-01-10T07:56:10.277000",
          "content": "<p>Hi Vincent, </p>\n\n<p>Thank you for reply. GPU was off. I turned it on. </p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 454440,
          "author_name": "Vincent Roseberry",
          "author_url": "",
          "post_date": "2019-01-11T16:38:10.813000",
          "content": "<p>Hi <a href=\"/huytran\">@huytran</a> &amp; <a href=\"/sezaugg\">@sezaugg</a>,</p>\n\n<p>Does making sure the GPU is turned ON for your kernel also fixes the problem for you?</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 452296,
      "author_name": "Abhishek Thakur",
      "author_url": "",
      "post_date": "2019-01-08T14:12:13.947000",
      "content": "<p>Cudnn and tensorflow versions mismatch.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 452260,
      "author_name": "Mamy Ratsimbazafy",
      "author_url": "",
      "post_date": "2019-01-08T13:09:14.503000",
      "content": "<p>This looks like a Tensorflow update.</p>",
      "votes": 1,
      "replies": []
    },
    {
      "id": 452380,
      "author_name": "Huy Tran",
      "author_url": "",
      "post_date": "2019-01-08T16:42:00.090000",
      "content": "<p>I got the same issue with my new kernels. The old ones are still working fine.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 452368,
      "author_name": "",
      "author_url": "",
      "post_date": "2019-01-08T16:20:11.617000",
      "content": "<p>Sometimes fork an run the new kernel magically solves things. </p>",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "452249": "Hey. \n\nI could not make a submission for the last three days. I run the exact same kernel that worked a week ago, but now I receive an error: \n\n... I could not publish the post with a problem, I dont know why. See attached picture ... \n      \nAlso, the time run of kernel increased from 6500s up to 15000s. I've heard that those problems might be a result of a massive kernels update. Is it so? Does anyone have a similar problem as I do? \n\nPS: I tried to post this message four times in a row, but only one was sucsessful. What is happening with kaggle? ",
    "453291": "Hi @DenisVorotyntsev\n\nFrom the logs you attached, it seems like you are using a Kernel with GPU disabled to train a CudnnRNN model which requires a GPU. \n\nCan you confirm that the GPU setting in the Kaggle Kernel Editor is turned on for you kernel?\n\nThank you",
    "452296": "Cudnn and tensorflow versions mismatch.",
    "452260": "This looks like a Tensorflow update.",
    "452380": "I got the same issue with my new kernels. The old ones are still working fine.",
    "452368": "Sometimes fork an run the new kernel magically solves things. "
  }
}